Meta's Secret Operation to Sabotage AI Rivals Exposed

Meta conducted a covert operation known internally as “Cannes,” utilizing an outside contracting firm called Covalen. This program involved third-party workers creating fake accounts impersonating minors under 18 to flood rival AI platforms, such as OpenAI’s ChatGPT, Google’s Gemini, and Character.AI, with inappropriate material.

Contractors were instructed to generate thousands of distressing prompts, focusing on themes like suicide and self-harm, with one phase tracking over 45,000 prompts. The project remained hidden from the targeted companies, raising questions about the data's use.

Contractors expressed discomfort over the disturbing nature of the prompts, fearing repercussions. Meta defended the operation as an “industry-standard practice” for safety benchmarking, but AI ethics experts criticized this justification, stating it masked anticompetitive behavior.

The scandal highlights a significant ethical breach, contrasting Meta's public stance on protecting teenagers on platforms like Instagram with its secretive actions against competitors.